More MARBL updates for CESM3 by mnlevy1981 · Pull Request #339 · ESCOMP/MOM_interface

mnlevy1981 · 2026-06-26T19:50:16Z

Need to set some grid-specific BGC parameters for the tx2_3v3 grid, and also NCAR/MOM6#434 changes how MOM6 expects to some BGC forcing data to be provided (so this PR updates MOM_input). There might be some other parameter changes coming from @kristenkrumhardt, she is tuning MARBL and I don't know if the MOM6 parameters should be set ParamGen or if we'll just update the default values in the Fortran code. I'll take this out of draft mode when all of that has been finalized.

Also add flags to support initializing MARBL tracers from file using native vertical grid (e.g. restart file). Note that currently the MARBL tracers are split across multiple restarts, so users will need to combine them into a single file before using this feature.

MOM6 now reads a single file containing two iron sediment fluxes and a vent flux, rather than one file per field.

Was not nesting conditionals correctly

I updated names of variables relating to the iron sediment and vent fluxes, which changes some defaults in MOM_input (and the input data file checker)

Hybrid cases default to MARBL_TRACERS_IC_FILE_IS_Z=False, and MARBL_IC_MIN_VAL is only used when initializing from Z

Also moved all aux_mom_MARBL tests to TL319_t233, made sure MARBL_IC_MIN_VAL is only set in tests that use MARBL, and created a new iron forcing file for tx2_3v2 with the old data presented in the new format (single file, different field names)

I had modified MOM_input.yaml but not input_data_list.yaml

mnlevy1981 · 2026-07-13T15:56:17Z

I'm running aux_mom on 1fa7196 right now, but splicing together the in-progress testing with results from before I added the MARBL conditional to setting MARBL_IC_MIN_VAL (which had a bunch of NLCOMP failures because I added MARBL_IC_MIN_VAL to every test instead of just the ones with MARBL) the failures are

4	FAIL
	2	NLCOMP (FAIL)
      1 SMS_Ld1_D.TL319_t232.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t232.G1850MARBL_JRA.derecho_intel
	2	BASELINE (FAIL)
      1 SMS_Ld1_D.TL319_t232.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t232.G1850MARBL_JRA.derecho_intel

These are expected; the NLCOMP failures come from the new MARBL_IC_MIN_VAL and the baseline failures are a combination of setting MARBL_IC_MIN_VAL (so the initial conditions in the tests are slightly different) and also the new tunings that are part of the updated MARBL tag.

I did also see

3	FAIL
	1	COMPARE (FAIL)
      1 PET.TL319_t232.C_JRA.derecho_intel
	1	TPUTCOMP (FAIL)
      1 ERI.TL319_t233.G_JRA.derecho_intel.mom-debug
	1	BASELINE (FAIL)
      1 PET.TL319_t232.C_JRA.derecho_intel

I'm not sure what happened with the PET test, so I'll see if it's reproducible before digging in... and my experience is that TPUTCOMP failures are intermittent / can be affected by machine and network use

mnlevy1981 · 2026-07-13T16:05:10Z

The PET test passes now:

$ cat TestStatus
PASS PET.TL319_t232.C_JRA.derecho_intel CREATE_NEWCASE
PASS PET.TL319_t232.C_JRA.derecho_intel XML
PASS PET.TL319_t232.C_JRA.derecho_intel SETUP
PASS PET.TL319_t232.C_JRA.derecho_intel SHAREDLIB_BUILD time=336
PASS PET.TL319_t232.C_JRA.derecho_intel NLCOMP
PASS PET.TL319_t232.C_JRA.derecho_intel MODEL_BUILD time=180
PASS PET.TL319_t232.C_JRA.derecho_intel SUBMIT
PASS PET.TL319_t232.C_JRA.derecho_intel RUN time=408
PASS PET.TL319_t232.C_JRA.derecho_intel COMPARE_base_single_thread
PASS PET.TL319_t232.C_JRA.derecho_intel BASELINE cime6.2.2-cesm3_cice6_6_3_16-dev_ncar_260625:
PASS PET.TL319_t232.C_JRA.derecho_intel MEMCOMP
PASS PET.TL319_t232.C_JRA.derecho_intel TPUTCOMP
PASS PET.TL319_t232.C_JRA.derecho_intel MEMLEAK
PASS PET.TL319_t232.C_JRA.derecho_intel SHORT_TERM_ARCHIVER

mnlevy1981 · 2026-07-13T19:56:49Z

I can confirm that only the expected tests are failing:

4	FAIL
	2	NLCOMP (FAIL)
      1 SMS_Ld1_D.TL319_t232.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t232.G1850MARBL_JRA.derecho_intel
	2	BASELINE (FAIL)
      1 SMS_Ld1_D.TL319_t232.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t232.G1850MARBL_JRA.derecho_intel

@alperaltuntas this is ready for review. Summary of changes (this is the intent of the changes, if anything goes against this list it's likely a mistake that I'm happy to fix):

testlist_mom.xml: move all aux_mom_MARBL tests from tx2_3v2 to tx2_3v3 (does not change pr_mom, aux_mom, prealpha, or prebeta test lists)
.gitmodules, MOM6/, and externals/MARBL update MOM6 to dev/ncar_260710 and MARBL to marbl0.50.0
MOM_input.yaml: updates to support MARBL compsets with tx2_3v3
- IC file was moved to grid_indpt since it's 1°x1° and MOM6 handles horizontal interpolation
- data from MARBL_FESEDFLUX_FILE, MARBL_FESEDFLUXRED_FILE, and MARBL_FEVENTFLUX_FILE were combined into single MARBL_FE_INTERIOR_SOURCE_FILE. There is a brand new file for tx2_3v3 and also I combined the data from the three existing tx2_3v2 files into a single file for that resolution (there are still a few tests with MARBL on the v2 grid in the test suite)
- Introduced MARBL_IC_MIN_VAL, which is necessary (but not sufficient) to pass DIMCSL test with MARBL compset
- Created river nutrient forcing files on v3 grid
input_data_list.yaml: mirrors changes to variables in MOM_input.yaml that deal with new forcing files (IC files in grid_indpt/, new iron forcing setup, new river nutrient files)

alperaltuntas · 2026-07-14T16:54:40Z

  <test name="ERS" grid="TL319_t232" compset="G1850MARBL_JRA">
    <machines>
-      <machine name="derecho" compiler="intel" category="aux_mom_MARBL"/>
      <machine name="derecho" compiler="intel" category="prebeta"/>


Can we also move this prebeta test to the newly added <test name="ERS" grid="TL319_t233" compset="G1850MARBL_JRA">, and remove this <test name="ERS" grid="TL319_t232" compset="G1850MARBL_JRA"> altogether?

Sure, that makes sense. That's done in 7a4148d

alperaltuntas · 2026-07-14T17:03:49Z

  url = https://github.com/NCAR/MOM6.git
  fxDONOTUSEurl = https://github.com/NCAR/MOM6.git
-    fxtag = dev/ncar_260624
+    fxtag = dev/ncar_260710


Could we update the tag to dev/ncar_260713 which include the latest changes from mom-ocean/main?

Done in 7a4148d

1. Moved prebeta test from tx2_3v2 -> tx2_3v3 (prealpha and aux_mom are still on tx2_3v2) 2. Updated to latest MOM6 tag (which includes merging latest main back in to dev/ncar)

mnlevy1981 · 2026-07-14T19:49:15Z

@alperaltuntas Looking at the testlist diffs now, there's one test where I moved an aux_mom_MARBL test from t232 to t233 but left the aux_mom and prebeta test on the old grid:

   <test name="SMS_Ld1_D" grid="TL319_t232" compset="C1850MARBL_JRA">
     <machines>
       <machine name="derecho" compiler="intel" category="aux_mom"/>
-      <machine name="derecho" compiler="intel" category="aux_mom_MARBL"/>
       <machine name="derecho" compiler="intel" category="prebeta"/>
     </machines>
   </test>
+  <test name="SMS_Ld1_D" grid="TL319_t233" compset="C1850MARBL_JRA">
+    <machines>
+      <machine name="derecho" compiler="intel" category="aux_mom_MARBL"/>
+    </machines>
+  </test>

Do you want me to move both of those tests to the new grid as well? I didn't want to change aux_mom tests before I had a chance to run them, but now that we see the expected NLCOMP and BASELINE failures I'm happy to move this to the new grid.

Similarly, there is one test in aux_mom and prebeta but not aux_mom_MARBL:

  <test name="SMS_Ld1_D" grid="TL319_t232" compset="C1850MARBL_JRA">
    <machines>
      <machine name="derecho" compiler="intel" category="aux_mom"/>
      <machine name="derecho" compiler="intel" category="prebeta"/>
    </machines>
  </test>

I think this should also be on the new grid.

All tests with MARBL active are now on the tx2_3v3 grid for all test suites

mnlevy1981 · 2026-07-14T21:28:57Z

Current plan is that I'll run the test suite one more time when derecho is back. I expect failures from the two tests that I moved to tx2_3v3 but the rest should still be bit-for-bit after moving to the latest MOM6 tag.

mnlevy1981 · 2026-07-15T15:45:01Z

The PET test is still in the queue, but the rest of the suite ran and results are as expected:

4	FAIL
	2	NLCOMP (FAIL)
      1 SMS_Ld1_D.TL319_t233.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t233.G1850MARBL_JRA.derecho_intel
	2	BASELINE (FAIL)
      1 SMS_Ld1_D.TL319_t233.C1850MARBL_JRA.derecho_intel
      1 SMS.TL319_t233.G1850MARBL_JRA.derecho_intel

These failures are now Missing baseline namelist and ERROR BFAIL baseline directory '<dir_name>' does not exist.

mnlevy1981 added 9 commits June 26, 2026 10:53

Update to newer iron forcing scheme

a584bd1

MOM6 now reads a single file containing two iron sediment fluxes and a vent flux, rather than one file per field.

Fix logic in input_data_list.yaml

b2c5c99

Was not nesting conditionals correctly

Variable names have changed

962395d

I updated names of variables relating to the iron sediment and vent fluxes, which changes some defaults in MOM_input (and the input data file checker)

Updated name of iron sed / vent flux file

63ba152

Merge tag 'mi_260625' into CESM3_MARBL_updates

f91da8f

Don't set MARBL_IC_MIN_VAL for hybrid

53236a0

Hybrid cases default to MARBL_TRACERS_IC_FILE_IS_Z=False, and MARBL_IC_MIN_VAL is only used when initializing from Z

Update MARBL and MOM6 tags

1fa7196

Also moved all aux_mom_MARBL tests to TL319_t233, made sure MARBL_IC_MIN_VAL is only set in tests that use MARBL, and created a new iron forcing file for tx2_3v2 with the old data presented in the new format (single file, different field names)

Update iron source file for tx2_3v2

a7ebb78

I had modified MOM_input.yaml but not input_data_list.yaml

mnlevy1981 marked this pull request as ready for review July 13, 2026 15:56

mnlevy1981 requested a review from alperaltuntas July 13, 2026 15:56

alperaltuntas reviewed Jul 14, 2026

View reviewed changes

Fixes from code review

7a4148d

1. Moved prebeta test from tx2_3v2 -> tx2_3v3 (prealpha and aux_mom are still on tx2_3v2) 2. Updated to latest MOM6 tag (which includes merging latest main back in to dev/ncar)

Move more tests to t233

c0afede

All tests with MARBL active are now on the tx2_3v3 grid for all test suites

mnlevy1981 merged commit 89f74c0 into ESCOMP:main Jul 15, 2026
6 checks passed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

More MARBL updates for CESM3#339

More MARBL updates for CESM3#339
mnlevy1981 merged 11 commits into
ESCOMP:mainfrom
mnlevy1981:CESM3_MARBL_updates

mnlevy1981 commented Jun 26, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

alperaltuntas Jul 14, 2026

Uh oh!

mnlevy1981 Jul 14, 2026

Uh oh!

alperaltuntas Jul 14, 2026

Uh oh!

mnlevy1981 Jul 14, 2026

Uh oh!

mnlevy1981 commented Jul 14, 2026

Uh oh!

mnlevy1981 commented Jul 14, 2026

Uh oh!

mnlevy1981 commented Jul 15, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

mnlevy1981 commented Jun 26, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

mnlevy1981 commented Jul 13, 2026

Uh oh!

alperaltuntas Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

mnlevy1981 Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

alperaltuntas Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

mnlevy1981 Jul 14, 2026

Choose a reason for hiding this comment

Uh oh!

mnlevy1981 commented Jul 14, 2026

Uh oh!

mnlevy1981 commented Jul 14, 2026

Uh oh!

mnlevy1981 commented Jul 15, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants